Podcast Episode 4

Episode 004 48 min

Wesley Belleman of California Air National Guard and Palo Alto Networks: Security Operations Center from Military and Private Industry Perspectives

with Wesley Belleman · California Air National Guard and Palo Alto Networks

0:00 / 48:00

Also on Spotify, Apple Podcasts, YouTube, Podbean, Player FM

Episode 4 of The Security Podcast of Silicon Valley, published June 17, 2021. Host Jon McLachlan talks with Wesley Belleman, then a cyber warfare operator in the California Air National Guard and a system engineer at Palo Alto Networks, after four years in the Air Force, about security operations: why he left the military, what a security operations center does, a SOC for every organization through an MSSP, Kasparov and automation, talking about space threats in public, security operations as a series of classification problems, five decision points, Security Onion and MITRE Caldera for hobbyists, buy when you can, the train track and the steering wheel, treating people like people, core values, the SOC accordion, Nowhere to Hide and SOAR, XDR, threat intelligence, and dwell time. About 48 minutes.

Episode notes

An interview with Wesley Belleman, Cyber Warfare Operator at the California Air National Guard and System Engineer at Palo Alto Networks, and his journey through Security Operations in the United Space Space Force.

In this episode, we dive into what a SOC is, how new technology changes our ability to respond to threats, and what the open problems are still open. We explore the cultural differences between civilian and military security operations and discuss various open-source projects such as Security Onion or MITRE Att&ck that can be used to build your own SOC. 

So what will the future of SOC's look like? Wes echo's Anton Chuvakin predictions that SOC's will be entirely automated, likely within our lifetimes, but not soon.

Hosted by Jon McLachlan — co-founder of YSecurity.

The deep dive

Nowhere to hide

Wesley Belleman, then a cyber warfare operator in the California Air National Guard and two months into Palo Alto Networks, with Jon on leaving the military, what a security operations center does, a SOC for every organization, Kasparov and automation, talking about threats in public, security operations as a series of classification problems, a hobbyist's stack of Security Onion and Caldera, buy when you can, the train track and the steering wheel, treating people like people, the SOC accordion, Nowhere to Hide, SOAR and XDR, threat intelligence, and shrinking dwell time.

Episode 4 of The Security Podcast of Silicon Valley aired on June 17, 2021, with Jon McLachlan hosting. The guest was Wesley Belleman, a cyber warfare operator in the California Air National Guard and a system engineer at Palo Alto Networks, after four years in the Air Force (Podbean, LinkedIn). The episode is called “Wesley Belleman of California Air National Guard and Palo Alto Networks: Security Operations Center from Military and Private Industry Perspectives.” It has no video version.

“And then the whole security operations process is just a series of classification problems.”

Wesley Belleman, at 15:20

Listen · Wesley Belleman on episode 4, from 14:59
About 25 seconds: Wes on why he treats security operations as a data problem. The clip stops on its own. Read the full transcript.

Wes served in the Air Force from May 2017 to May 2021. His LinkedIn says he owned the requirements for Kraken, a Space Force system for real-time defense against cyber threats, and led a team at United States Southern Command. He joined Palo Alto Networks in May 2021 and the California Air National Guard the same month, part-time (LinkedIn). Jon says on the tape that Wes has been at Palo Alto Networks for two months.

Four months before the episode, Wes published “Classification in Security Operations” in Towards Data Science, the article he brings up on the tape. It breaks a SOC’s work into five decisions, each a classification problem (Towards Data Science).

This page follows the conversation in order: a section for each part of it with the sources beside it, fifteen figures, thirteen clips from the tape, the full transcript, and five tools: a quiz, a scorecard for your security operations, a calculator for the analyst hours that alert triage takes, a timeline, and a to-do list. Where the decodes of the tape disagree on a word, the transcript puts its best guess in brackets, and the page doesn’t quote those words.

A debt of gratitude

Jon asks for the story behind Wes’s service. Wes says he felt “a debt of gratitude to the military,” so he wanted to join. Once he did, the training and education it gave him helped kickstart his career straight out of college.

He says he has written on LinkedIn about problems with the military’s personnel practices, and that he doesn’t think the military will stay competitive with Silicon Valley. He says that isn’t to bash the military or the people in it. He doesn’t want others to leave for the same reasons he did.

“I want people to stay in, especially talented people, to continue to go to the military and to stay there and build out their careers there and be successful.”

Wesley Belleman, at 2:21

A timeline. May 2017: Air Force, project manager, to May 2021. He owned the requirements for the Space Force's Kraken, for real-time defense against cyber threats, and led a team at United States Southern Command. February 2021: the article, Classification in Security Operations, in Towards Data Science. May 2021, filled: he joins Palo Alto Networks and starts part-time in the California Air National Guard. June 2021: this episode, two months into Palo Alto Networks. March 2023: lead cyber planner, with Cyber Dawn, Amber Mist and a NATO exercise. April 2026, filled: commander of a 44-person Cyber Protection Team in U.S. Cyber Command's Cyber Mission Force. Wes on why he joined: "a debt of gratitude to the military" Wes, at 1:42
Wes's path from the Air Force to a Cyber Protection Team, from his LinkedIn.

Up or out

Jon asks why he left. Wes says the military still has archaic HR practices. You get pigeonholed into one career, you get little say in where you live, and moving sideways is hard.

“So you kind of get into a very specific career and you get pigeonholed into that career, and they kind of tell you where you’re going to move next.”

Wesley Belleman, at 2:38

Then there’s up or out. Wes says books on cyber defense for the Department of Defense talk about it, such as The Perfect Weapon. David E. Sanger of The New York Times published The Perfect Weapon: War, Sabotage, and Fear in the Cyber Age in 2018 (Wikipedia). People who want to stay hands-on as engineers get pushed into management.

“So the up or out is a problem for some people who want to be more technical.”

Wesley Belleman, at 3:00

Jon expects the military to change, slowly.

“You know, it’s like the larger the organization, the slower the rate of change.”

Jon McLachlan, at 3:29

He adds that this is why startups can be interesting: they change quickly.

What a SOC does

Jon asks Wes to explain a security operations center to listeners who are new to it. Wes says that when he joined the Air Force, he didn’t realize how much of a niche security operations could be, and he found he had to explain it to a lot of people.

His answer starts from monitoring. Logs and network traffic all go somewhere. The question is whether anyone reads them, compares them against signatures of known attacks, and follows up when something matches.

“And they’re actually looking at those logs and they’re actually following those alerts and they’re following up to them”

Wesley Belleman, at 5:27

If they find that an attack worked, they do incident response.

The future, he says, is closing that loop faster: see the alert, investigate, decide whether it had an impact, and respond before the impact grows. Wikipedia describes a SOC as responsible for protecting an organization against threats, with analysts who monitor the network and investigate incidents (Wikipedia).

Listen · Wesley Belleman on episode 4, from 4:42
About 40 seconds: Wes on the difference between logging and security operations. The clip stops on its own. Read the full transcript.
Five boxes in a row with arrows: logs and traffic go somewhere. Compare them to signatures. An alert when something matches. Investigate: did it work? Respond, filled: limit the impact. Below, three cards. Logging alone: logs and network traffic pile up somewhere. The question is whether anybody reads them. Security operations: analysts triage the logs and traffic, follow up on each alert, and do incident response when an attack worked. The future, dashed: close that loop faster, so you respond before the impact on your network is significant, or at least keep it small. Wes's question: "And is anybody really reading it?" Wes, at 4:52
What a SOC does with an alert, from Wes's answer.

Police officers inside the network

Jon offers a comparison: the front line, like Navy SEALs, but for cyber warfare. Wes says that’s a good way of looking at it, especially on the defensive side, since Navy SEALs are more offensive. He compares a SOC to your security forces, or your “police officers within your network,” there to keep everybody safe.

Every organization, but how?

Jon asks which organizations should have security operations. Every organization, Wes says. The question is how. A big organization, such as Walmart, can run its own SOC in-house and pay for the analysts and engineers.

“But if you’re a small organization, then you probably want to outsource it to, like, an MSSP.”

Wesley Belleman, at 6:56

A managed security service provider, or MSSP, is a company that provides outsourced network security services (Wikipedia). Wes’s examples are Robert Herjavec’s Herjavec Group and big providers such as IBM. Herjavec, a panelist on Shark Tank, founded The Herjavec Group in 2003 (Wikipedia). Wes says an MSSP cuts costs because it centralizes experts who serve several different tenants. Even for a small company, he would look into one, so a security operations center monitors every endpoint and workload.

Two cards. A big organization: runs its own security operations center in-house, and pays the high cost of good analysts and engineers. Wes's example is Walmart. A small organization, filled: outsources to a managed security service provider, or MSSP. One MSSP centralizes experts who serve several different tenants, which cuts costs. Below, three points. Wes's examples of MSSPs: the Herjavec Group, which Robert Herjavec of Shark Tank founded in 2003, and big providers such as IBM. A brand-new organization might not be thinking about it yet, or might not be able to afford it. Filled: even for a small company, Wes would look into an MSSP, so a SOC monitors every endpoint and workload. Wes's line: "But if you're a small organization, then you probably want to outsource it to, like, an MSSP." Wes, at 6:56
Running a SOC in-house or through an MSSP, from Wes's answer.

Kasparov and the SOC

Jon wonders whether most of this could be automated one day. Wes brings up a book Garry Kasparov wrote about chess and computers, which says the future is a mix of the human and the computer. Kasparov introduced advanced chess, also called centaur chess, in which each human player uses a computer chess engine (Wikipedia). Wes says that may already be untrue in chess, where a human now adds little to a computer, and that the level of complexity in security operations is different.

“from what I’ve seen, that’s really far off.”

Wesley Belleman, at 9:11

“A lot of security operations centers still have yet to implement automation.”

Wesley Belleman, at 9:13

Many SOCs, he says, are still working on getting all their data in one place. He expects a different conversation in five to ten years.

Talking about threats in public

Jon asks what threats Wes saw at the Space Force.

“And as you might imagine, I’m not able to talk about specific threats to space systems.”

Wesley Belleman, at 10:05

He says he saw a push for the government to talk more about threats. Generals who want funding or new priorities need voters who understand why. His example is General Raymond, then Chief of Space Operations, speaking openly about a Chinese satellite with a robotic arm that can grab other satellites. John W. Raymond was the first chief of space operations, from 2019 to 2022 (Wikipedia).

“And he was trying to show people this is why we need to think about defending our satellites.”

Wesley Belleman, at 11:27

Wes hopes more information on cyber threats opens up over the next five to ten years, so voters take it into account at the voting booth.

Listen · Wesley Belleman on episode 4, from 10:57
About 35 seconds: Wes on why the Space Force started talking about threats in public. The clip stops on its own. Read the full transcript.

Physical access means compromise

Jon ties the satellite story to a rule of his own.

“whenever you have a physical compromise of a system, then all of the secrets and everything inside of it, you have to consider them compromised as well.”

Jon McLachlan, at 12:31

He adds that a well-informed public matters for a free society.

A data problem

Jon asks what role data plays in attack detection and response.

“I think the data is starting to become the primary basis upon which people think about probably security overall.”

Wesley Belleman, at 13:01

Wes says books on security operations from the last couple of years start with data engineering and data science concepts, such as what it means for an analytic to be sensitive versus specific. Then they expand into security operations and its data sources.

“So I’ve seen different ways of breaking it up, but the one that I kind of like is network data sources, host data sources and service data sources.”

Wesley Belleman, at 13:49

The three sources flow into a repository, usually some sort of NoSQL database, and the analytics run on that data. Wes says a lot of people still think of it as a security problem. With a background in intrusion detection or prevention systems, they want to write their traditional rules. An intrusion detection system is a device or software application that monitors a network or systems for malicious activity or policy violations (Wikipedia). Friends in data science showed Wes the data view from the outside, and he brought it into his work.

“If you’re trying to find malicious traffic as compared to benign traffic, that’s just a binary classification problem within data science.”

Wesley Belleman, at 15:00

Split the malicious traffic further, into command and control or exfiltration, and each split is a classification problem too.

Three boxes with arrows down: network data sources, host data sources and service data sources. They flow into a repository, usually some sort of NoSQL database, and then into analytics, filled: is an analytic sensitive, or specific? Below, three cards. A binary classification: malicious traffic or benign traffic. More categories: command and control, exfiltration and others. Each split is a classification problem too. The older habit, dashed: writing traditional rules, from a background in intrusion detection or prevention systems. Wes's line: "If you're trying to find malicious traffic as compared to benign traffic, that's just a binary classification problem within data science." Wes, at 15:00
Three kinds of data sources in one repository, from Wes's answer.

Five decision points

Jon says that sounds cut out for machine learning.

“So I don’t know if you saw my article on this, but I had kind of five decision points”

Wesley Belleman, at 15:41

He defined each one as a classification problem.

The five decisions in his article are whether data is useful for detection, whether an event is an incident, whether the context data is sufficient, whether it’s a true incident, and how well the mitigation worked. The article defines a cybersecurity incident as “an event that an analyst decided to investigate further” (Towards Data Science). Wes says the article gives his subjective view of how machine learning fits each decision point. Some are a good fit, and a big one that’s becoming common is spam filters.

“And spam filters obviously has a huge impact in the phishing space.”

Wesley Belleman, at 16:04

Natural language processing can judge whether an email looks like phishing. You can’t catch everything, he says, but keeping more phishing emails out of your network makes it less likely that someone exploits it through one. That’s one of the early decision points in his paper: is this an attack, or traffic you can keep allowing through?

Listen · Wesley Belleman on episode 4, from 15:58
About 40 seconds: Wes on spam filters as a classification problem. The clip stops on its own. Read the full transcript.
Five decisions down a line. Decision 1, useful data? Is this data useful for detection? Decision 2, filled, an incident? Is this an incident? A probability threshold decides, and alert fatigue is the risk. Decision 3, enough context? Is the context data sufficient? Adding context is data hydration. Decision 4, a true incident? False positives get closed. Decision 5, did it work? How effective was the mitigation? The article expects machine learning to add little here. Two cards. The article's definition: a cybersecurity incident is an event that an analyst decided to investigate further. On the tape, dashed: spam filters. Natural language processing judges whether an email looks like phishing, and keeps more of it out of the network. Wes's line: "So I don't know if you saw my article on this, but I had kind of five decision points" Wes, at 15:41
The five decisions in Wes's article, each a classification problem.

Jon says the show will link the article in the description, and the show notes do.

Collaboration across branches

Jon asks whether the branches of the military, and the three-letter agencies, work together on security operations.

“We don’t want to be wasting a lot of resources by duplicating effort.”

Wesley Belleman, at 17:50

Wes says they talked to every branch and to certain agencies, to see what they could pull from each other. The Space Force came out of the Air Force, so most of his collaboration was with the Air Force. Wikipedia says the Space Force was established on December 20, 2019, and is organized as part of the Department of the Air Force (Wikipedia). The collaboration has a cost.

“Sometimes you’re in, like, 10 meetings a day.”

Wesley Belleman, at 18:30

As long as you can manage that alongside building your SOC, he says, the collaboration is important.

Security Onion for hobbyists

Jon asks which tools, open source projects or services help process the data and flag what’s suspicious. Wes starts with the project most people in the security operations community use to get started. He says it’s used more by hobbyists than by big organizations.

“So the Security Onion framework comes with two network sensors, so you can start collecting your network data.”

Wesley Belleman, at 19:45

The sensors are Suricata and Zeek. They feed the Elastic stack, a NoSQL database you view with Kibana. Elastic also has an endpoint log collector, which Wes calls Log Beats, and you point it back at the same Elastic stack. Then your endpoint logs, service logs and network data sit in one place, where you can run queries, build signatures and generate analytics.

“You can kind of detect, like, a whole life cycle of an attack.”

Wesley Belleman, at 20:47

Security Onion is a free Linux distribution for threat hunting, security monitoring and log management, developed by Doug Burks in 2008 (Wikipedia). Vern Paxson began work on Zeek in 1995 (Wikipedia). Suricata, from the Open Information Security Foundation, had its first standard release in July 2010 (Wikipedia). Elasticsearch is a search engine developed by Elastic (Wikipedia), and Kibana is a dashboard for it (Wikipedia). The show notes link Security Onion Solutions.

Listen · Wesley Belleman on episode 4, from 19:45
About 35 seconds: Wes on the pieces of Security Onion. The clip stops on its own. Read the full transcript.
Three boxes with arrows down: Suricata, a network sensor. Zeek, a network sensor. Log Beats, for endpoint logs. They feed the Elastic stack, filled, with network, endpoint and service data in one place, viewed in Kibana for queries, signatures, analytics and dashboards. Then test it: four boxes in a row. A Caldera agent, dashed, a simulated attacker. ATT&CK tactics: pivot, command and control, exfiltrate. Security Onion: what did it see? A detection, filled, in Kibana. Wes's line: "You can kind of detect, like, a whole life cycle of an attack." Wes, at 20:47
A hobbyist's Security Onion stack, and a Caldera test against it, from Wes's answers.

Test it with Caldera

To validate your own security tools, Wes points to Caldera, an open source project from MITRE built on the MITRE ATT&CK matrix. Jon admits he hasn’t used any of these tools. Caldera has campaigns you run on your own endpoints. You install its agent on an endpoint.

“So you just simulate, okay, bad guy got a zero-day and was able to get some sort of an agent on your endpoint.”

Wesley Belleman, at 21:15

The agent runs tactics from the ATT&CK framework, such as pivoting from one endpoint to another, command and control from a server, or exfiltrating data. Then you look in Security Onion to see what that looks like, and write detections that show up in your Kibana dashboard. ATT&CK is a guideline for classifying and describing cyberattacks and intrusions, created by the Mitre Corporation and released in 2013 (Wikipedia). The show notes link MITRE ATT&CK.

Listen · Wesley Belleman on episode 4, from 21:12
About 35 seconds: Wes on testing detections with Caldera. The clip stops on its own. Read the full transcript.

Buy when you can

Jon asks what it takes to move from a hobby setup to something production ready at a large company. Wes says build versus buy was a huge topic and concern of his at the Space Force. A big company can probably afford enough engineers to build everything, but he doesn’t think that’s generally the most efficient way.

“I’m generally, as much as possible, a proponent of buy when you can.”

Wesley Belleman, at 22:45

His example is Elastic. Download Security Onion and you manage your own Elastic cluster. Elastic is also a company, with a cloud-provided Elasticsearch engine that it manages, and professional services. Splunk has similar offerings, he says, and tends to cost more than Elastic.

“at a high level, you get what you pay for.”

Wesley Belleman, at 23:38

Run it yourself and you configure log rotation yourself, for example to keep logs for 30 days. Miss that and logs sit for years until your storage blows up, or one index holds too many and your memory does.

Listen · Wesley Belleman on episode 4, from 23:35
About 40 seconds: Wes on what hosted software saves you. The clip stops on its own. Read the full transcript.

Jon is for buy over build almost every time. He calls himself “the expert of very little in this world,” and a managed service lets him use other people’s expertise without having to worry about it.

“It almost seems like a no-brainer.”

Jon McLachlan, at 24:55

He’s surprised when companies spend a lot of time and money building things that already exist. He wants to spend the time on open technical problems that haven’t been solved yet. Wes agrees.

Two cards. Run Elastic yourself: download Security Onion and you manage your own Elastic cluster, including log rotation, storage and the size of each index. Buy it hosted, filled: Elastic Cloud or Elastic's professional services, run by people who spent hours learning how to keep it up. Splunk has similar offerings and tends to cost more. What goes wrong when you run it yourself: logs you meant to keep for 30 days sit there for years. Your storage blows up. Too many logs land in one index, and your memory blows up. Wes's line: "at a high level, you get what you pay for." Wes, at 23:38
Running Elastic yourself or buying it hosted, from Wes's answer.

Two months at Palo Alto Networks

Jon asks about the biggest challenge in the private sector so far, two months in. He says he’d heard people who leave the military describe a cutthroat world of personal performance, and he hasn’t seen that at Palo Alto Networks. He says “it’s very team focused,” and it feels like people care about each other.

The biggest issue for him, he says, was a feeling of instability on leaving government work, which can be more stable on salaries and through the economy. Jon tells him good for him for taking the risk, and adds that you can always go back if it doesn’t work out.

Freedom to turn left or right

Jon asks what the private sector does better. Wes says less security comes with a lot of freedom. He can live where he wants, as long as he can get a job there. He can move laterally, from system engineer to software developer, if he builds the skills and someone will hire him.

In the military, he says, you get on a train track and have to go forward on that track.

“your control is the pedal, right?”

Wesley Belleman, at 29:30

The pedal sets how fast you go along the track, and you can’t turn.

“Versus in the private sector, you can turn left or right if you want to.”

Wesley Belleman, at 29:36

Listen · Wesley Belleman on episode 4, from 29:18
About 20 seconds: Wes on career paths in the military and the private sector. The clip stops on its own. Read the full transcript.

Jon says career paths at large companies have two tracks, and one is the individual contributor, or IC: the engineers and the operations folks.

“And you can climb the ladder just as well as an IC sticking to engineering, and you become an architect.”

Jon McLachlan, at 30:09

Two cards. The military: you get on a train track and go forward on it. Your control is the pedal: how fast you go along the track. You can't turn left or right. The private sector, filled: live where you want, if you can get a job there. Move laterally, say from system engineer to software developer, if someone will hire you. Below: up or out, people who want to stay hands-on get pushed into management. Jon: large companies have two tracks, and you can climb as an individual contributor and become an architect. Wes's line: "Versus in the private sector, you can turn left or right if you want to." Wes, at 29:36
The train track and the steering wheel, from Wes's answer.

Treat people like people

Jon asks what the military does culturally that Wes would like to see more of in the private sector. Wes says the military is as diverse as the private sector, and it depends on where you go. Service branches organize, train and equip, he explains. Combatant commands do the war fighting. A unified combatant command is a joint command of the Department of Defense made up of units from two or more service branches (Wikipedia), and United States Southern Command is one of eleven (Wikipedia). Wes says the work there was more operational.

“everybody felt like a person, like everybody treated me like a person.”

Wesley Belleman, at 32:24

People weren’t there to do their work and go home, he says. They were there to grow relationships and help each other out. He has friends at companies where people punch in, do as much work as possible and punch out, and he says parts of the military are like that too.

“every company should just realize that everyone you work with is a person.”

Wesley Belleman, at 33:05

He says to treat them like a person, be friendly with them, and be kind to them. He sees that attitude at Palo Alto Networks, and says other companies don’t have that mentality.

Listen · Wesley Belleman on episode 4, from 32:50
About 30 seconds: Wes on treating the people you work with like people. The clip stops on its own. Read the full transcript.

Celebrate disagreement

Jon agrees.

“I think it’s super important to see people first as people.”

Jon McLachlan, at 33:36

He likes to assume two things about the people he interacts with. They bring something to the table that’s different from what he brings, and they want to do the right thing. So a disagreement means one of two things is about to happen. Someone learns something new about how the world works, or someone sees the world from a new perspective where the problem goes away.

“acknowledge, like, our differences, not as baggage”

Jon McLachlan, at 35:36

We all have different life experiences, he says, and we should see those differences as strengths.

Wes likes the phrase celebrate disagreement, from an article of Jon’s. Some leaders see disagreement as a reason they can’t decide, he says.

“Sometimes disagreement is just part of the process.”

Wesley Belleman, at 36:46

Two boxes at the top with arrows down: they bring something different to the table, and they want to do the right thing. Both lead to a filled box: a disagreement, where one of two things is about to happen. Two boxes below: someone learns how the world works, or someone sees it from a new perspective. Below: Jon has been on both the giving and the receiving side, and says it makes everything enjoyable. Wes likes the phrase celebrate disagreement. Some leaders see disagreement as a reason they can't decide. Wes's line: "Sometimes disagreement is just part of the process." Wes, at 36:46
What a disagreement leads to, from Jon's answer.

Core values

Wes compares two sets of core values.

“One thing I like about Palo Alto is one of our core values is disruption.”

Wesley Belleman, at 36:53

The Air Force’s core values are integrity first, service before self, and excellence in all we do. Wes calls them great core values that can feel me-centric. Palo Alto Networks’ core values, he says, include disruption, collaboration and inclusion. Disruption is about changing the whole system. The Space Force started with the Air Force’s three, temporarily, and is setting up its own.

Jon expects larger organizations to change slower, and sees an upside for the Space Force.

“Maybe that’s actually a benefit for the Space Force, because it’s new, maybe a little bit smaller.”

Jon McLachlan, at 38:02

Two cards. The Air Force: integrity first. Service before self. Excellence in all we do. Great core values, Wes says, that can feel me-centric. Palo Alto Networks, filled: disruption, collaboration and inclusion, in Wes's words. Disruption is about changing the whole system. Below: the Space Force started with the Air Force's three, temporarily, and is setting up its own. Jon: a new, smaller service might be able to innovate more quickly than the larger branches. Wes's line: "One thing I like about Palo Alto is one of our core values is disruption." Wes, at 36:53
The core values Wes compares.

The SOC accordion

Jon asks for predictions about the SOC and attack detection. Wes points to Anton Chuvakin, who he says works at Google and was at Gartner for many years, and whose articles and research he has learned from. Chuvakin, Wes says, seems to predict that the SOC could be fully automated at some point. Wes remembers the range Chuvakin gave.

“it might be 2025, it might be 2045.”

Wesley Belleman, at 39:16

The show notes link Chuvakin’s LinkedIn. In a post cross-posted to CISO Platform on March 9, 2021, Chuvakin argues that SOCs can’t and shouldn’t be fully automated today. Then he lays out scenarios that could change that, such as attackers who automate faster, and asks whether his advice still holds in 2025, 2030 or 2035. He gives no firm date (Anton on Security, CISO Platform). Wikipedia lists Chuvakin at Google Cloud’s Office of the CISO, and before that as a research vice president at Gartner (Wikipedia). At Gartner in 2013, he coined the term endpoint threat detection and response (Wikipedia).

Wes says Chuvakin has swayed him a little in that direction. He also expects a lot of change as companies collect more data and learn which data sources are useful.

“I could see the size of SOCs kind of ballooning and then shrinking back and then growing, kind of up and down, like a little accordion.”

Wesley Belleman, at 39:50

A SOC might collect every endpoint log it can think of and add an endpoint protection tool, from Symantec, Palo Alto Networks or another company. Then it finds that a lot of that data doesn’t have much predictive power for attack detection.

Listen · Wesley Belleman on episode 4, from 39:50
About 35 seconds: Wes on why he expects SOCs to grow and shrink. The clip stops on its own. Read the full transcript.
Three boxes with arrows: balloon, collect every log. Shrink, filled, drop what doesn't help. Grow again, up and down. Two cards. Why it shrinks: a SOC collects every endpoint log it can think of and adds endpoint protection tools, then finds much of the data has little predictive power for attack detection. Chuvakin's question, dashed: in March 2021 he argued SOCs can't and shouldn't be fully automated today, and asked if that still holds in 2025, 2030 or 2035. No firm date. Wes's line: "I could see the size of SOCs kind of ballooning and then shrinking back and then growing, kind of up and down, like a little accordion." Wes, at 39:50
Why Wes expects the SOC to grow and shrink.

Nowhere to Hide

Wes gave a presentation at the Space Force several times called Nowhere to Hide, on security automation. Jon asks whether it’s spelled K-N-O-W or N-O. Wes says N-O, like nowhere. Jon sums it up: you can’t run, and there’s nowhere to hide.

Wes compares it with encryption, since Jon runs an encryption company. Done right, encryption means an adversary who can see your data can’t read it. Nowhere to hide is about what the adversary does.

“if the adversary is doing something that they’re not supposed to be doing, they can’t do that undetected, right?”

Wesley Belleman, at 41:18

They get caught and stopped quickly, he says, and security automation is a huge part of that. SOAR platforms such as Phantom and Palo Alto’s Cortex XSOAR have hundreds of integrations. A SOC with dozens of tools, and tool sprawl, can stop logging in to each one.

“They can bring it all together in a SOAR platform, and then they can start to add playbooks, which are just basically scripts.”

Wesley Belleman, at 41:57

For people who don’t like scripting, a playbook can be visual programming. As automation grows, SOC operations get faster, and operators move out of tier one work, such as opening tickets and closing obvious false positives, toward more complicated threat detection. Security orchestration, automation and response, or SOAR, is a group of technologies that let organizations respond to some incidents automatically (Wikipedia). Phantom was one of a series of companies Splunk acquired between 2013 and 2019 (Wikipedia).

Listen · Wesley Belleman on episode 4, from 41:15
About 20 seconds: Wes on the idea behind his Nowhere to Hide presentation. The clip stops on its own. Read the full transcript.
Three boxes with arrows: dozens of tools and tool sprawl. A SOAR platform, filled, with hundreds of integrations. Playbooks, as scripts or visual programming. Two cards. Tier one, automated, dashed: opening tickets and closing obvious false positives. Analysts move on, filled: to more complicated threat detection, while SOC operations get faster. Below: Wes's examples are Phantom, and Palo Alto's Cortex XSOAR. The idea of his talk, Nowhere to Hide: an adversary doing what they're not supposed to do can't do it undetected. Wes's line: "They can bring it all together in a SOAR platform, and then they can start to add playbooks, which are just basically scripts." Wes, at 41:57
What SOAR changes in a SOC, from Wes's answer.

XDR

The other trend Wes expects in the near future is in XDR.

“So XDR is an interesting concept.”

Wesley Belleman, at 42:45

Extended detection and response, he says, does detection and response across the endpoints, the network and other devices, where endpoint detection and response or network detection and response each work within one. Some people could argue it has existed as long as security operations. He counts data loss prevention as part of it. A lot of companies use Red Hat or Windows, so their logs look alike, and most network traffic is probably TCP/IP.

“So if you can correlate Windows logs with TCP/IP, then those analytics that correlate those two different data sources can be shared across different organizations.”

Wesley Belleman, at 43:41

Vendors keep proprietary analytics in their cloud. Customers send their data there, and the analytics send a report back to the customer’s SOC on traffic that looks suspicious. Wes calls the selling of analytics one of the major new trends in the market, along with threat intelligence. Nir Zuk of Palo Alto Networks coined the term XDR in 2018 (Wikipedia).

Listen · Wesley Belleman on episode 4, from 43:50
About 30 seconds: Wes on analytics shared across organizations. The clip stops on its own. Read the full transcript.
Three boxes with arrows down: endpoints, such as Windows and Red Hat. The network, mostly TCP/IP. More sources, such as other devices and data loss prevention. They meet in a filled box: extended detection and response, with detections across the sources in a single data object. Then three boxes in a row: your data, sent to the vendor. The vendor cloud, with proprietary analytics. A report, filled, back to your SOC. Below: many companies run the same systems, so analytics that correlate Windows logs with TCP/IP can be shared across organizations. Wes counts the selling of analytics, and of threat intelligence, among the major new trends. Wes's line: "So XDR is an interesting concept." Wes, at 42:45
Detection in one stream and across them, from Wes's answer.

Threat intelligence

Jon asks whether threat intelligence is the fingerprints that identify specific types of attacks. Wes says yes. An adversary who attacks Palo Alto’s network will want to use that attack on several other networks, to see how many they can compromise.

“Like, ransomware works best if you can hit as many targets as possible and get them all to pay.”

Wesley Belleman, at 44:54

If Palo Alto Networks detects a ransomware attack, and stops it before anything happens, it can share what that attack looks like, so others can make sure it doesn’t hit their environment.

Dwell time

Jon pictures a ransomware crew breaking in, finding the storage and identifying all the backups. That sounds targeted to him, and hard to automate. Wes says some ransomware attacks have been automated, and names the time Jon describes.

“That’s called the dwell time.”

Wesley Belleman, at 45:42

It runs from when the adversary is in your network to when they’re detected. Security automation and more sensors aim to shrink it. His example is SolarWinds. He says it started around June or July 2020 and was detected by CrowdStrike in December 2020, about five or six months of dwell time. Wikipedia says the attack was first publicly reported on December 13, 2020, that the hackers had access for eight to nine months, and that FireEye says it discovered the attack while investigating its own breach (Wikipedia).

Wes also brings up a CrowdStrike report on adversaries’ breakout times, the time you have to detect an adversary to stop them. The fastest adversary on it, he says, had a breakout time of about two hours.

The goal, he says, is to get dwell time down from six months in some cases.

“right down to something that’s measured in hours or minutes.”

Wesley Belleman, at 46:52

Listen · Wesley Belleman on episode 4, from 45:41
About 20 seconds: Wes on dwell time. The clip stops on its own. Read the full transcript.
Three boxes with arrows: in your network, the adversary gets in. Dwell time, filled: nobody has seen it. Detected: dwell time ends. Three numbers. 5 to 6 months: SolarWinds dwell time, as Wes tells it on the tape. 8 to 9 months: how long the SolarWinds hackers had access, per Wikipedia. About 2 hours: the fastest breakout time in the CrowdStrike report Wes cites. Below: Wes credits CrowdStrike with the detection. Wikipedia says FireEye found the attack while investigating its own breach, and that it was first reported on December 13, 2020. Filled: the goal is dwell time measured in hours or minutes, with security automation and more sensors. Wes's line: "right down to something that's measured in hours or minutes." Wes, at 46:52
From months to hours, from Wes's answer and Wikipedia.

More automated, faster, more secure

Jon closes on the future.

“It will be more automated. It’ll be faster.”

Jon McLachlan, at 47:04

It will be more secure too, in no small part because of people like Wes on the front lines. Jon says they’ll have Wes back to see how the predictions work out.

After the tape

Wes stayed at Palo Alto Networks. He is now a cyber security consultant there, a specialist for customers across North America, and he helped write an exam for its new certification program (LinkedIn).

From cyber warfare operator to commander. In the California Air National Guard, Wes was a cyber security manager until February 2025, and he has been a lead cyber planner since March 2023. In April 2026 he became the commander of a 44-person Cyber Protection Team, part of U.S. Cyber Command’s Cyber Mission Force (LinkedIn). U.S. Cyber Command was established on June 23, 2009, and is a unified combatant command (Wikipedia). Since April 2026 he has also been a planner in the 195th Wing’s plans and programs office. The wing’s space units moved to the U.S. Space Force in 2025 (Wikipedia).

Exercises with other organizations. As lead cyber planner, Wes worked on Cyber Dawn 2024 and 2025, CalGuard’s first Amber Mist, and a NATO exercise where his team placed first of eight. His LinkedIn puts Cyber Dawn 2025 at “363 participants across 52 organizations” (LinkedIn).

Kraken in the news. On July 22, 2022, Washington Technology reported that the Space Force’s cyber operations division planned a seven-year cybersecurity services contract called Digital Bloodhound to protect its ground networks. It named two product lines. Manticore duplicates and analyzes data, and Kraken gives more real-time defense against active threats (Washington Technology). Wes’s LinkedIn says he owned the requirements for Kraken.

Splunk became part of Cisco. Cisco agreed to buy Splunk for $28 billion in September 2023 and completed the deal on March 18, 2024 (Wikipedia).

More on this show. On episode 2, a month before this one, Andrew Gontarczyk, then the CISO of Pure Storage, calls the blue team the security operations center. For security data, listen to episode 50 with Alan Braithwaite of RunReveal on SIEM, and episode 63 with Kabir Mathur of Leen on why security tools don’t talk to each other. For threat intelligence, listen to episode 12 with Dan Sherry and Grace Chi of Pulsedive. On episode 90, Garrett Smith of Reveal Technology talks about building for the military.

A timeline. June 17, 2021: this episode, Wes two months into Palo Alto Networks. July 22, 2022: Digital Bloodhound. Washington Technology reports the Space Force's contract plan, with product lines named Manticore and Kraken. March 2023: lead cyber planner in the California Air National Guard, with Cyber Dawn, Amber Mist and a NATO exercise. March 18, 2024: Cisco completes its $28 billion purchase of Splunk. 2025: the 195th Wing's space units go to the U.S. Space Force. 2025: Cyber Dawn 2025, 363 participants across 52 organizations, in Wes's LinkedIn. April 2026, filled: Wes takes command of a 44-person Cyber Protection Team.
What happened after the episode, from LinkedIn, Washington Technology and Wikipedia.

Did you catch it? A seven-question quiz

Seven questions from the episode. Each answer links the moment in the transcript.

Quiz · seven questions

Did you catch it?

Pick an answer to see whether you caught it, and where it is on the tape.

  1. What does Wes say the whole security operations process is?

    1. A series of classification problems
    2. A set of compliance checklists
    3. A job for one analyst per network

    a. "And then the whole security operations process is just a series of classification problems." Transcript, 15:20.

  2. Which two network sensors come with Security Onion, according to Wes?

    1. Snort and Wireshark
    2. Suricata and Zeek
    3. Kibana and Splunk

    b. "So that's Suricata and Zeek." Transcript, 19:51.

  3. What does Wes suggest for validating your own security tools?

    1. A yearly bug bounty
    2. A vendor's demo environment
    3. MITRE's open source project Caldera

    c. "And then to validate your own security tools, MITRE has a really cool open source project called Caldera." Transcript, 20:52.

  4. Where does Wes stand on build versus buy?

    1. Buy when you can
    2. Always build it yourself
    3. Build first, buy later

    a. "I'm generally, as much as possible, a proponent of buy when you can." Transcript, 22:45.

  5. On Wes's military train track, what do you control?

    1. The steering wheel
    2. The pedal
    3. The next station

    b. "your control is the pedal, right?" Transcript, 29:30.

  6. What did Wes call the presentation he gave at the Space Force several times?

    1. Defense in Depth
    2. Assume Breach
    3. Nowhere to Hide

    c. "So I have a presentation I gave in the Space Force several times called Nowhere to Hide." Transcript, 40:39.

  7. What does Wes call the time from an adversary getting into your network to being detected?

    1. Dwell time
    2. Breakout time
    3. Patch window

    a. "That's called the dwell time." Transcript, 45:42.

Is there nowhere to hide on your network?

Wes’s answers add up to a checklist for security operations. Eight questions on how much of it you have today.

Scorecard · eight questions

Is there nowhere to hide on your network?

Answer for the systems that matter most. If two answers fit, pick the lower one.

  1. Does someone read your alerts and follow up on them every day?

    • No
    • When there's time
    • Yes, with a named owner

    "And is anybody really reading it?" See What a SOC does.

  2. Do your network, host and service data land in one place?

    • No
    • Two of the three
    • All three

    "They're still just kind of grappling their hands around getting all the data in one place." See Kasparov and the SOC.

  3. Who watches your endpoints and workloads?

    • Nobody
    • Our team, in business hours
    • A SOC, in-house or an MSSP

    "But if you're a small organization, then you probably want to outsource it to, like, an MSSP." See Every organization, but how?

  4. Have you run a simulated attacker against your own detections?

    • Never
    • Once
    • On a schedule

    "And then to validate your own security tools, MITRE has a really cool open source project called Caldera." See Test it with Caldera.

  5. Do you know how long you keep your logs, and who manages that?

    • No
    • Roughly
    • Yes, set and owned

    "because you might only want your logs to stay there for 30 days." See Buy when you can.

  6. Are the routine triage steps automated with playbooks?

    • None
    • A few
    • Most tier one work

    "and then they can start to add playbooks, which are just basically scripts." See Nowhere to Hide.

  7. Do you run detections across endpoints, the network and other sources together?

    • No
    • Partly
    • Yes

    "And so bringing all those data sources together" See XDR.

  8. Do you know the dwell time of your last incident?

    • No
    • Estimated
    • Measured

    "is to shrink that dwell time as much as possible." See Dwell time.

Easy to hide

An attacker could sit in your network for months. Start by naming who reads the alerts, and get your network, host and service data into one place.

Partly watched

The data is there. Next, test your detections with a simulated attacker and automate the triage steps your team repeats most.

Nowhere to hide

You watch, test and automate. Keep measuring dwell time, and add the data sources that prove useful.

How many analyst hours does triage take?

Wes says playbooks move analysts out of tier one work, such as closing obvious false positives. This calculator shows how many analyst hours your alerts take each day, and how many automation gives back.

Calculator

How many analyst hours does triage take?

Three numbers: alerts a day, minutes to triage one, and the share a playbook handles.

24.0analyst hours a day
16.0hours playbooks give back a day
3.0analysts on eight-hour shifts

With 400 alerts a day at 6 minutes each, triage takes 24.0 analyst hours a day, or 3.0 analysts on eight-hour shifts. Playbooks give back 16.0 hours a day.

The 400 alerts, 6 minutes and 40% are our example. Put in your own numbers.

From Zeek in 1995 to a Cyber Protection Team in 2026

The tools Wes names, the ideas behind them, his path, and the events around them. Filter by topic.

Timeline

From Zeek in 1995 to a Cyber Protection Team in 2026

  1. Zeek

    Vern Paxson begins work on the network analysis framework (Wikipedia, on this show).

  2. Security Onion

    Doug Burks develops the Linux distribution for threat hunting and security monitoring (Wikipedia, on this show).

  3. U.S. Cyber Command

    Established as a sub-unified command (Wikipedia).

  4. Suricata

    The first standard release, from the Open Information Security Foundation (Wikipedia, on this show).

  5. ATT&CK

    The Mitre Corporation releases its guideline for classifying cyberattacks and intrusions (Wikipedia, on this show).

  6. Endpoint detection

    Anton Chuvakin of Gartner coins endpoint threat detection and response (Wikipedia).

  7. Air Force

    Wes starts as a project manager, through May 2021 (LinkedIn).

  8. XDR

    Nir Zuk of Palo Alto Networks coins the term (Wikipedia, on this show).

  9. The Perfect Weapon

    David E. Sanger publishes The Perfect Weapon: War, Sabotage, and Fear in the Cyber Age (Wikipedia, on this show).

  10. Space Force

    The service is established, as part of the Department of the Air Force (Wikipedia, on this show).

  11. SolarWinds

    The attack is first publicly reported (Wikipedia, on this show).

  12. Classification in Security Operations

    Wes's article breaks a SOC into five classification decisions (Towards Data Science, on this show).

  13. Humans in the SOC

    Anton Chuvakin argues SOCs can't be fully automated today, and asks whether that holds in 2025, 2030 or 2035 (CISO Platform, on this show).

  14. Palo Alto Networks

    Wes joins Palo Alto Networks and the California Air National Guard (LinkedIn).

  15. Episode 2

    Andrew Gontarczyk calls the blue team the security operations center (episode 2).

  16. Episode 4

    Wes on the show, two months into Palo Alto Networks (on this show).

  17. Digital Bloodhound

    Washington Technology reports the Space Force contract plan, with Manticore and Kraken (Washington Technology).

  18. Lead cyber planner

    Wes's role in the California Air National Guard, with Cyber Dawn and a NATO exercise (LinkedIn).

  19. Cisco and Splunk

    Cisco completes its $28 billion purchase of Splunk (Wikipedia, on this show).

  20. Space units move

    The 195th Wing's space units move to the U.S. Space Force (Wikipedia).

  21. Cyber Dawn 2025

    363 participants across 52 organizations, in Wes's LinkedIn (LinkedIn).

  22. Episode 90

    Garrett Smith of Reveal Technology on building for the military (episode 90).

  23. Cyber Protection Team

    Wes takes command of a 44-person team in U.S. Cyber Command's Cyber Mission Force (LinkedIn).

What to do this week

Five things, in the order of the episode.

  1. Name who reads the alerts. Write down who follows up on each kind of alert today, and when. Take the scorecard with your team. Our Monitoring and Response team can do the watching if nobody has the job.
  2. Put the data in one place. List your network, host and service data sources and where each one lands. Check how long you keep the logs and who manages that.
  3. Test one detection. Run one ATT&CK technique on a test endpoint with Caldera and check that an alert fires. A penetration test shows what a real attacker would try.
  4. Count your triage hours. Run the calculator with your own alert numbers, then write playbooks for the steps your analysts repeat most.
  5. Get a second pair of eyes. A vCISO can own the plan, and you can book 15 minutes with us.

Key takeaways

  1. Every organization needs security operations

    Wes says every organization should have security operations. A big company can run its own SOC, and a small one can outsource to a managed security service provider.

  2. Someone has to read the logs

    Logs and traffic pile up somewhere. Security operations means someone triages them, follows up on each alert, and responds when an attack worked.

  3. Treat detection as a data problem

    Network, host and service data go into one repository. Telling malicious traffic from benign traffic is a binary classification problem.

  4. A series of classification problems

    Wes's article splits security operations into five decisions, from whether data is useful for detection to how well a mitigation worked.

  5. Start with Security Onion and Caldera

    Security Onion bundles Suricata, Zeek and the Elastic stack for a hobbyist's setup. MITRE's Caldera runs ATT&CK techniques against it so you can build detections.

  6. Buy when you can

    Running your own Elastic cluster means handling log rotation, storage and memory yourself. Wes prefers hosted software and professional services.

  7. Automate tier one work

    SOAR platforms pull dozens of tools together with playbooks, so analysts move from closing obvious false positives to harder threat detection.

  8. Correlate across data sources

    XDR runs detections across endpoints, the network and other devices, and vendors sell analytics that work across organizations.

  9. Shrink the dwell time

    Dwell time runs from when an adversary gets in to when they're detected. Wes wants it measured in hours or minutes, down from months.

  10. Treat people like people

    Wes saw it at United States Southern Command and sees it at Palo Alto Networks. Jon adds that a disagreement ends with someone learning something or seeing a new perspective.

Chapters

Tap a chapter to play from there.

Pull quotes

“a debt of gratitude to the military”

Wesley Belleman ·

“I want people to stay in, especially talented people, to continue to go to the military and to stay there and build out their careers there and be successful.”

Wesley Belleman ·

“So you kind of get into a very specific career and you get pigeonholed into that career, and they kind of tell you where you're going to move next.”

Wesley Belleman ·

“So the up or out is a problem for some people who want to be more technical.”

Wesley Belleman ·

“You know, it's like the larger the organization, the slower the rate of change.”

Jon McLachlan ·

“And is anybody really reading it?”

Wesley Belleman ·

“And they're actually looking at those logs and they're actually following those alerts and they're following up to them”

Wesley Belleman ·

“But if you're a small organization, then you probably want to outsource it to, like, an MSSP.”

Wesley Belleman ·

“from what I've seen, that's really far off.”

Wesley Belleman ·

“A lot of security operations centers still have yet to implement automation.”

Wesley Belleman ·

“And as you might imagine, I'm not able to talk about specific threats to space systems.”

Wesley Belleman ·

“And he was trying to show people this is why we need to think about defending our satellites.”

Wesley Belleman ·

“whenever you have a physical compromise of a system, then all of the secrets and everything inside of it, you have to consider them compromised as well.”

Jon McLachlan ·

“I think the data is starting to become the primary basis upon which people think about probably security overall.”

Wesley Belleman ·

“So I've seen different ways of breaking it up, but the one that I kind of like is network data sources, host data sources and service data sources.”

Wesley Belleman ·

“If you're trying to find malicious traffic as compared to benign traffic, that's just a binary classification problem within data science.”

Wesley Belleman ·

“And then the whole security operations process is just a series of classification problems.”

Wesley Belleman ·

“So I don't know if you saw my article on this, but I had kind of five decision points”

Wesley Belleman ·

“And spam filters obviously has a huge impact in the phishing space.”

Wesley Belleman ·

“We don't want to be wasting a lot of resources by duplicating effort.”

Wesley Belleman ·

“Sometimes you're in, like, 10 meetings a day.”

Wesley Belleman ·

“So the Security Onion framework comes with two network sensors, so you can start collecting your network data.”

Wesley Belleman ·

“You can kind of detect, like, a whole life cycle of an attack.”

Wesley Belleman ·

“So you just simulate, okay, bad guy got a zero-day and was able to get some sort of an agent on your endpoint.”

Wesley Belleman ·

“I'm generally, as much as possible, a proponent of buy when you can.”

Wesley Belleman ·

“at a high level, you get what you pay for.”

Wesley Belleman ·

“It almost seems like a no-brainer.”

Jon McLachlan ·

“your control is the pedal, right?”

Wesley Belleman ·

“Versus in the private sector, you can turn left or right if you want to.”

Wesley Belleman ·

“And you can climb the ladder just as well as an IC sticking to engineering, and you become an architect.”

Jon McLachlan ·

“everybody felt like a person, like everybody treated me like a person.”

Wesley Belleman ·

“every company should just realize that everyone you work with is a person.”

Wesley Belleman ·

“I think it's super important to see people first as people.”

Jon McLachlan ·

“acknowledge, like, our differences, not as baggage”

Jon McLachlan ·

“Sometimes disagreement is just part of the process.”

Wesley Belleman ·

“One thing I like about Palo Alto is one of our core values is disruption.”

Wesley Belleman ·

“Maybe that's actually a benefit for the Space Force, because it's new, maybe a little bit smaller.”

Jon McLachlan ·

“it might be 2025, it might be 2045.”

Wesley Belleman ·

“I could see the size of SOCs kind of ballooning and then shrinking back and then growing, kind of up and down, like a little accordion.”

Wesley Belleman ·

“if the adversary is doing something that they're not supposed to be doing, they can't do that undetected, right?”

Wesley Belleman ·

“They can bring it all together in a SOAR platform, and then they can start to add playbooks, which are just basically scripts.”

Wesley Belleman ·

“So XDR is an interesting concept.”

Wesley Belleman ·

“So if you can correlate Windows logs with TCP/IP, then those analytics that correlate those two different data sources can be shared across different organizations.”

Wesley Belleman ·

“Like, ransomware works best if you can hit as many targets as possible and get them all to pay.”

Wesley Belleman ·

“That's called the dwell time.”

Wesley Belleman ·

“right down to something that's measured in hours or minutes.”

Wesley Belleman ·

“It will be more automated. It'll be faster.”

Jon McLachlan ·

About the guest

Wesley Belleman, Cyber Security Consultant, Palo Alto Networks

Wesley Belleman is a cyber security consultant at Palo Alto Networks, which he joined in May 2021, and he helped write an exam for its new certification program. In the California Air National Guard he was a cyber security manager from May 2021 to February 2025, has been a lead cyber planner since March 2023, and became commander of a 44-person Cyber Protection Team in April 2026. From 2017 to 2021 he was a project manager in the Air Force, where he owned the requirements for the Space Force's Kraken and led a team at United States Southern Command. He wrote Classification in Security Operations for Towards Data Science in February 2021. He joined Jon McLachlan on episode 4, published June 17, 2021.

About the host

Jon McLachlan is co-founder of YSecurity, the on-demand cybersecurity team for startups, and of Cyberbase.ai. The Security Podcast of Silicon Valley is hosted by Jon McLachlan and Sasha Sinkevich. Jon McLachlan on LinkedIn ↗

Context and sources

  • Episode 4 on Podbean, June 17, 2021, with show notes on Wes's path through security operations in the Space Force, what a SOC is, his Towards Data Science article, Security Onion, MITRE ATT&CK and the future of SOCs. Source ↗
  • Wesley Belleman's LinkedIn profile. Source ↗
  • Wesley Belleman's experience on LinkedIn: the Air Force from May 2017 to May 2021, Palo Alto Networks from May 2021, and the California Air National Guard from May 2021, including command of a Cyber Protection Team from April 2026. Source ↗
  • Wesley Belleman, Classification in Security Operations: Targeted machine learning to defend your network, Towards Data Science, February 11, 2021. Source ↗
  • Washington Technology, Space Force wants a 'bloodhound' to sniff out cyberthreats, by Nick Wakeman, July 22, 2022: Digital Bloodhound, Manticore and Kraken. Source ↗
  • Anton Chuvakin, Stop Trying to Take Humans Out of SOC … Except … Wait… Wait… Wait…, Anton on Security. Source ↗
  • The same post by Anton Chuvakin, cross-posted on CISO Platform, March 9, 2021. Source ↗
  • Anton Chuvakin on LinkedIn, linked from the show notes. Source ↗
  • Security Onion Solutions, linked from the show notes. Source ↗
  • MITRE ATT&CK, linked from the show notes. Source ↗
  • Wikipedia on the security operations center. Source ↗
  • Wikipedia on managed security services and managed security service providers. Source ↗
  • Wikipedia on Robert Herjavec, who founded The Herjavec Group in 2003. Source ↗
  • Wikipedia on advanced chess, introduced by Garry Kasparov. Source ↗
  • Wikipedia on John W. Raymond, the first chief of space operations. Source ↗
  • Wikipedia on David E. Sanger, author of The Perfect Weapon: War, Sabotage, and Fear in the Cyber Age (2018). Source ↗
  • Wikipedia on intrusion detection systems. Source ↗
  • Wikipedia on Security Onion, developed by Doug Burks in 2008. Source ↗
  • Wikipedia on Zeek, which Vern Paxson began in 1995. Source ↗
  • Wikipedia on Suricata, from the Open Information Security Foundation. Source ↗
  • Wikipedia on Elasticsearch. Source ↗
  • Wikipedia on Kibana. Source ↗
  • Wikipedia on ATT&CK, released by the Mitre Corporation in 2013. Source ↗
  • Wikipedia on the United States Space Force, established December 20, 2019, as part of the Department of the Air Force. Source ↗
  • Wikipedia on unified combatant commands. Source ↗
  • Wikipedia on United States Southern Command. Source ↗
  • Wikipedia on Anton Chuvakin. Source ↗
  • Wikipedia on endpoint detection and response, a term Anton Chuvakin coined at Gartner in 2013. Source ↗
  • Wikipedia on security orchestration, automation and response. Source ↗
  • Wikipedia on Splunk, including Phantom and Cisco's purchase, completed March 18, 2024. Source ↗
  • Wikipedia on extended detection and response, a term Nir Zuk of Palo Alto Networks coined in 2018. Source ↗
  • Wikipedia on the 2020 United States federal government data breach, the SolarWinds attack. Source ↗
  • Wikipedia on United States Cyber Command. Source ↗
  • Wikipedia on the 195th Wing of the California Air National Guard. Source ↗

Transcript

Automatic transcription. Jon McLachlan hosts this episode with Wesley Belleman, then a cyber warfare operator in the California Air National Guard and a systems engineer at Palo Alto Networks. There is no diarization on the tape, so speaker labels were assigned from the content and checked against reference recordings of each voice from this episode and by pitch, and a few short lines may sit with the wrong speaker. Names and product names were normalized where the recognizer misheard them (Jon McLachlan, Wesley Belleman, SOC, Palo Alto Networks, Robert Herjavec, the Herjavec Group, Garry Kasparov, Anton Chuvakin, Security Onion, Suricata, Zeek, Kibana, MITRE ATT&CK, Caldera, Cortex XSOAR, CrowdStrike, SolarWinds), as were a few terms (SOAR, XDR, DLP, MSSP, NoSQL, TCP/IP, dwell time). Words in [brackets] are best guesses where the audio is unclear, and [unclear] marks a word we couldn't make out. Answers are transcribed as spoken, with repeated words and false starts trimmed. Short backchannels such as "yeah" and "right" are left out where they overlap the other speaker.

Read the full transcript 9,028 words

Jon McLachlan

Welcome, everyone, to the Security Podcast of Silicon Valley. I'm here today with a very special guest, Wesley Belleman, who is the cyber warfare operator for the California National Guard and a system engineer at Palo Alto Networks. He's spent a good chunk of time in the armed services. So thank you very much for your service, Wes. [This was] started off at the Air Force. You were a cybersecurity engineer at the Air Force. Then you moved over to the Space Force, which was a new branch of the military. And you were at the United States Southern Command for a while as a technical lead. And now you've cut over to private industry with your new gig at Palo Alto Networks as a security engineer. And of course you continue to serve in the California Air National Guard as a cyber warfare operator. So thank you so much for joining me on the show. Impressive background. A huge thanks for all of that service.

Wesley Belleman

Yeah, no, thank you for having me. I'm definitely looking forward to talking about security operations.

Jon McLachlan

So clearly, you're a very talented guy, and talented people always have options available to them. And so I'm super curious, is there a story behind all of your service in the military and the armed forces?

Wesley Belleman

No, yeah, that's a great question. Well, first of all, my parents were in the military, so there was the family business. After my dad left the military, he still continued to work for the Department of Defense as a civilian. Actually, a teacher. And I had the opportunity to live on different military bases. So I definitely felt that the military had given me a lot, from a home, a career for my dad, and school was even funded by the DoD. So I definitely felt like a debt of gratitude to the military. So I really wanted to join. And then when I did join, there was just a lot of benefits. They help kind of kickstart your career by giving you training and education. So there was just a lot of enticing things about joining straight out of college that made it exciting for me. So I'm definitely really glad I joined the military. You know, I talked a little bit on LinkedIn about some issues I have with their personnel practices and how I think the military is not going to remain competitive in the future with Silicon Valley, but that's certainly not to bash on the military or the people in it. That's just because I don't want people to leave for the same reasons I did. I want people to stay in, especially talented people, to continue to go to the military and to stay there and build out their careers there and be successful.

Jon McLachlan

I'm super curious now, what was the reason that you left the military?

Wesley Belleman

No, that's great. So yeah, the military has, like, they still have some kind of archaic HR practices. So you kind of get into a very specific career and you get pigeonholed into that career, and they kind of tell you where you're going to move next. You don't really have a lot of choice in where you live. You don't have a lot of choice if you want to move laterally. There's also a lot of up or out, so that's talked about in some books on cyber defense for the DoD, such as The Perfect Weapon. He talks about that. So the up or out is a problem for some people who want to be more technical. You know, if you're the type of person that you like to be hands-on, you like to solve problems, you like to be an engineer, then you're quickly pushed into management and you can't really move forward and become a senior engineer, a software engineer. 10 years is very valuable outside the military, but in the military, it's hard to do that.

Jon McLachlan

I know those are some interesting questions that you raise, and I'm sure that they'll change. It's just a slower process for it to change. And constantly changing for the better. You know, it's like the larger the organization, the slower the rate of change. It's true almost across the board, right? And that's why startups can be very interesting, because they change very, very quickly. But if you want a little bit more stability, like a larger company, I think people sort of discover that and realize that, even in the private sector. So you actually spend a good chunk of time as a cybersecurity engineer focus on operations and security operations. And so you worked in a SOC, right? A security operations center. So do you want to describe for the audience, who's maybe interested in security but not familiar with a SOC, what is a SOC, exactly?

Wesley Belleman

Right, no, that's a good question. There's probably a lot of people, I actually, when I joined the Air Force, I didn't realize how much of a kind of a niche within security, security operations actually could be. I think because there's a kind of a heavy emphasis on security operations within the military, as far as the defensive side overall and the cybersecurity side overall. But I did find that I needed to explain to a lot of people what is the security operations center? What are security operations? And really, it's similar to monitoring. So I know you had a guest on and you talked about monitoring. A lot of people ask, are you doing monitoring? Are you doing logging? But when you're just doing logging or you're just watching your network traffic, that's all going somewhere. And is anybody really reading it? Are you really triaging those different logs and the different network traffic, trying to compare it against a signature, which is a known attack format that you can compare it to, and is someone actually looking at that? And when that does alert, and you say, okay, well, we've received this traffic coming into our network, and we know that it matches an attack, are we looking to see where did that come from? Was it successful? And so security operators sit in a security operations center, and it can be tied to a single network or to many networks. And they're actually looking at those logs and they're actually following those alerts and they're following up to them, potentially doing incident response if they find that there actually was some sort of a successful attack. And then the future, which we can talk about, of that is being able to just close that loop and make it a lot faster, so that once you actually see that alert, you're able to investigate it, determine if it had an impact, and then respond to it before it really had a significant impact on your network, or at least be able to minimize that impact. So that's really what security operations are and how it ties to monitoring, which I think maybe is a concept that all security people will be more familiar with.

Jon McLachlan

So it's kind of like the front line, the Navy SEALs, but for cyber warfare, right?

Wesley Belleman

Yeah, no, that's a good way of looking at it, especially on the defensive side. I mean, I guess Navy SEALs are more like offensive, but a little bit offensive, yeah. But right, it's very much like your security forces or your police officers within your network that are there trying to just keep everybody safe within the network.

Jon McLachlan

So what sort of organization would you recommend have security operations?

Wesley Belleman

Yeah. So I would say, I mean, maybe I'm biased because of my background, but I would say every organization should have that within their cybersecurity portfolio. The question is more going to be how, right? So if you're a Walmart or a really big organization, you probably want to have your own security operations center that you run in-house, and you can fork over the high cost of a good security operations analyst and an engineer like myself who can help set up a security operations center. But if you're a small organization, then you probably want to outsource it to, like, an MSSP. So this is an emerging concept, the managed security service provider. So for those who watch Shark Tank, Robert Herjavec from Shark Tank has his own MSSP, the Herjavec Group. So that's one example. And then, of course, there's the big ones out there, like IBM and [AT&T], which also have their own MSSP offerings. And they can really help smaller organizations just monitor all their own endpoints and networks without having to have kind of, like, security analyst personnel. And that also helps cut down costs, because they're able to centralize experts within one MSSP that serves several different tenants. So my recommendation, again, it all comes down to cost and where you're at in your organization. Like, if you're a brand new organization, you might just not be thinking about it, might not be able to afford it. But even for a small company, I would definitely consider looking into an MSSP, just to make sure that all of your endpoints and all of your workloads are being monitored by a security operations center.

Jon McLachlan

Yeah, no, that's spectacular. It makes me wonder if one day we'll get to a point where some of that stuff is, or most of that stuff, could be automated. But the human element is often important and necessary when hunting down live, unfolding security events, right?

Wesley Belleman

Right. So this is something, this is, yeah, this is a pretty hot topic right now. And I don't know if you've heard of, Garry Kasparov wrote a book about automated chess and how there's kind of the mix of the human and the computer is the future, and you kind of need both. And that book was pretty relevant, and now it seems that that actually might already not be true. So when he was originally writing that book, it would be, you know, a human and a computer could be a computer, like, with 10 times the capacity of chess. And now it no longer seems that that's true, that basically, like, a human is now just minimal help with a computer at playing chess. We're just irrelevant now. And chess is not the same thing as security operations. The level of complexity is very different, but there's an analogy there, right? Maybe that will happen in a lot of other different fields, such as security operations. I definitely think that's, you know, from what I've seen, that's really far off. A lot of security operations centers still have yet to implement automation. They're still just kind of grappling their hands around getting all the data in one place. So there's a lot of work to do before we get to the point where it's Skynet. But as we move in that direction, I think 10 years, or five to 10 years from now, we'll be having a different conversation. But right now it's very, very difficult to see that horizon of everything being automated, because there's so much work to do. There's so much work to do to even just start doing the basics of automation, to start thinking of using automation to really cut out your analysts.

Jon McLachlan

No, sounds like a great opportunity for some entrepreneurial thinking. It'll be exciting to watch that unfold, you know. So with Space Force and your service in the military, what sort of threats did you actually see or deal with? Can you get into any of the details?

Wesley Belleman

Yeah, no, I definitely appreciate that question. And as you might imagine, I'm not able to talk about specific threats to space systems. That's something that, while I was at Space Force, it was great to see some of that opening up, in the sense that there was definitely a push to have the government start talking more about threats to our systems and threats to the country overall. There's kind of the traditional mentality that a lot of these things need to be secret, that any kind of intercontinental or international conflict needs to be very under the weeds. But what people started realizing is that as people are getting more politically involved, people are more informed through the internet. And then when generals are trying to get additional funding or trying to change national priorities, they're not able to do that, because they don't have supportive constituents, because voters don't understand why are you doing this. And to give a non-cyber example of that, General Raymond, who's the commander of the U.S. Space Force, the Chief of Space Operations, he started talking openly about a Chinese satellite that has, maybe you've heard of this, that has, like, a robotic arm on it. So he started talking openly about that. And he wanted people to know, like, there is a satellite out there from a foreign government that is able to grapple other satellites, because it has this robotic arm. And he was trying to show people this is why we need to think about defending our satellites. And then obviously you have to understand additionally the importance of GPS and our day-to-day lives, or the importance of satellite communications in the military overall. But if you can kind of understand that, and then he can speak openly about this satellite, then that starts to paint a picture for voters and constituents and people why this is important, why we're doing this. And there's a lot of people trying to do the same thing on the cyber side, but it's like you said, just some things take longer in the government. So hopefully over the next, you know, the next five to 10 years, we'll start to see more information open up so people can start getting involved. You know, when they're thinking about security for their government, especially cybersecurity, they can know, you know, why are we doing this? Why is this a priority? And so I really hope that, you know, voters will start to think about this and will start to politically take this into account when they go to the voting booth, right?

Jon McLachlan

Yeah, absolutely. Yeah, that's super interesting about the Chinese satellite with the arm. You know, it bumps into that rule that I have: whenever you have a physical compromise of a system, then all of the secrets and everything inside of it, you have to consider them compromised as well. I like the role of a well-informed democratic society. It seems very important to maintaining a free society. So when you're in there and you're in the weeds with security operations, what role does data really play in attack detection and your response?

Wesley Belleman

No, that's such an important question right now. I think the data is starting to become the primary basis upon which people think about probably security overall. I mean, you can tell me, you're on the encryption side, and you've had some other roles, but definitely on the security operations side as well. I've noticed I've started reading more books that are released in the last couple of years on security operations, and they definitely start with data. They start with data engineering and data science concepts, and then they expand from there into the security operations field. So they say, okay, you need to understand what is an analytic, what does it mean for an analytic to be sensitive versus specific, et cetera. And then they talk about different data sources within security operations. So I've seen different ways of breaking it up, but the one that I kind of like is network data sources, host data sources and service data sources. And so you have those three data sources coming into usually some sort of a NoSQL database, but into whatever kind of repository you want it to be in. And then from there, you do analytics on that data. And thinking of it from that perspective, and starting from that just basic data organization, data analysis mindset, is not necessarily new or modern per se, but I just see more and more people joining that chorus of, we need to think about this as a data problem. Like, I know a lot of people are still thinking of it as a security problem. Maybe they have kind of this background with IDSs or IPSs, and they just want to write their traditional rules like they were before. But more people are starting from data and then moving towards security operations. And it took me a while to do that. I was fortunate that I had some friends who were in the data science field, and so they were able to show that to me from the outside, and then I was able to bring that into my toolkit within my work. But I definitely think that, you know, thinking of it as a data problem, I mean, at the end of the day, right? If you're trying to find malicious traffic as compared to benign traffic, that's just a binary classification problem within data science. So if you're thinking about it from there, and then as you try to break it up into further categories, whether it's command and control traffic or something like that, or exfiltration, then you just realize it's just a classification problem. And then the whole security operations process is just a series of classification problems.

Jon McLachlan

So you turn it into, like, a machine learning classification problem. I mean, I'm listening to you, and I'm thinking, wow, that just cut itself right out for machine learning algorithms to just chew in and try to classify all of that data.

Wesley Belleman

Yes. So I don't know if you saw my article on this, but I had kind of five decision points, and each of them was, I defined as a classification problem. And I gave my really subjective view about how I felt machine learning fit into those different decision points. For some of them, I think it's a really good fit, and we're going to see a lot of growth. I think a big one, a really big one that's becoming kind of common, is just spam filters. And spam filters obviously has a huge impact in the phishing space. So if you're just trying to keep phishing out of your network, you can also just block phishing emails, which can be detected with machine learning, right? It can do natural language processing to kind of determine if that email looks like it's a phishing email or it's a regular [benign] email. And obviously you can't catch everything, but if you can at least keep more phishing emails out of your perimeter, out of your network, then you're just less likely to have someone exploit your network through a phishing email. So that's just one example of one of the early decision points that I have in my paper about, you know, just being able to determine, like, is this really an attack, or is this fine? That's some, like, traffic that you can kind of just continue to allow through.

Jon McLachlan

Yeah, no, we'll put a nice link to that in the description of the podcast, so people can go off and look it up and dig into the details there. But that's a nice overview. So in terms of the data and the sources of the data, I know you were specifically focused on probably Space Force for most of your career in the services. But did you find that there was collaboration between the different branches of the military, or even the three-letter agencies? Or was that maybe an opportunity to have a wider, more global view of, you know, the US government's systems' security response, and you try to unify some of that security operations?

Wesley Belleman

Right, no, absolutely. So, you know, again, sometimes I complain about the DoD, but we do try our best to collaborate, just because we care about the money that comes from the citizens and just how we prioritize resources. We don't want to be wasting a lot of resources by duplicating effort. So we do try to talk to each other and see, okay, where can we collaborate? What have you done that we can pull from? So we did a good job. I think we talked to every branch of the [DoD], and including, you know, we do talk to certain agencies, as you can imagine. And definitely, you know, the Space Force, rolled out of the Air Force, is still part of the Department of the Air Force. So probably most of my collaboration was with the Air Force, just because of how close they were, for lack of a better term. But we did a really good job of talking to all the other different agencies and branches. So sometimes the collaboration can bog you down. Sometimes you're in, like, 10 meetings a day. But as long as you can kind of manage that with, I have to get something done, I have to build my SOC, then we definitely do a lot of that. And I think that's really important as well. Just like I was talking about earlier, people try to understand how they see the government. The government belongs to them. Additional transparency is just collaboration and good use of government resources.

Jon McLachlan

Yeah, no, definitely. I'm really impressed and happy to hear that, actually. So in terms of actually where the rubber hits the road, would you recommend any specific tooling? Or are there open source projects that help process all of that influx of data and are specifically designed to identify threats or suspicious traffic? Or maybe there's even services out there that you could subscribe to, and they'll consume your logs and flag suspicious things on your behalf, or?

Wesley Belleman

Yes, absolutely. So the most popular project that we use in the security operations community just to get started, and this is more used just by hobbyists, it's not really used by, like, big organizations, is the Security Onion framework. So the Security Onion framework comes with two network sensors, so you can start collecting your network data. So that's Suricata and Zeek. And then those feed into the Elastic stack, which again is a NoSQL database that you can view with Kibana. And then Elastic also has an endpoint log collector called Log Beats. And so, yeah, I would recommend it. If someone wants to tinker around and kind of be a hobbyist, I would recommend to find the Security Onion framework, and then install Log Beats on some endpoints to monitor those logs. And then you can point those directly back to your Elastic stack. And then you've got your log data, and your endpoint log data, your service log data, and your network data, all in your same Elastic stack. And so then you can start running queries, you can build signatures, you can start generating analytics, you can have the Elastic stack create new data dictionaries to correlate across those three data sensors, which I think is, you know, kind of where things start to get really interesting. You can kind of detect, like, a whole life cycle of an attack. So that'd be my recommendation. And then to validate your own security tools, MITRE has a really cool open source project called Caldera. I don't know if you've heard of that.

Jon McLachlan

But I have to admit, I haven't used any of these great tools.

Wesley Belleman

Yeah, so Caldera is really great, and it's based on the MITRE ATT&CK matrix. And they have several different campaigns that you can run on your endpoints. So you basically install the agent on your endpoint. So you just simulate, okay, bad guy got a zero-day and was able to get some sort of an agent on your endpoint. And then they start running some of those tactics that are in the MITRE ATT&CK framework, such as pivoting from one endpoint to another, or command and control from a server into that endpoint, exfiltrating data from that endpoint. And then if you've done the Security Onion framework, then you can go see in the Security Onion framework, what does that actually look like? And then you can start developing detections for that, to actually have those populate in your Kibana dashboard.

Jon McLachlan

So if I'm an engineer at a large company, and I'm tasked with thinking about how do I detect some of these attacks coming in, and I want to start to get serious. You mentioned the Onion framework as, you know, it's great if I want to do some hobby stuff, but what sort of commitment is required to move that to the next level and really have it be production ready? Is it just a time commitment to see it through all the way to the end, or are there scaling issues that you wouldn't want to even start with that project? Or what would you recommend?

Wesley Belleman

No, it's been a huge topic and concern of mine when I was in the Space Force, the build versus buy paradigm. And I mean, you can do it either way. Like, if you're at a big company, you can probably afford enough engineers to build everything. But I personally wouldn't say that that's generally the most efficient way at solving that issue. I'm generally, as much as possible, a proponent of buy when you can. And so [for instance], for example, Elastic, right? So Elastic is in the Security Onion framework. But if you download Security Onion, you're going to have to manage your own Elastic cluster there. And Elastic itself, like, Elastic is a company. So they have a cloud-provided Elasticsearch engine, which they manage, and, you know, smart people there spent several hours learning about how Elastic works and how to keep it up and running, et cetera. And so my suggestion to someone at that company would be to consider using Elastic Cloud, or using their professional services to try to maintain your Elastic cluster. And obviously, you know, Splunk, which depends on your environment, but that tends to be a little more expensive than Elastic, has similar offerings. And so generally, it's just, you know, at a high level, you get what you pay for. So I would recommend investing more money in better software, hosted software when possible, because then you're going to make sure that things don't go down, that your logs are rotating properly, because you might only want your logs to stay there for 30 days. And if you manage it yourself, you're going to have to configure that on your own, and you might have logs sitting for years, and then your storage is going to blow up, or you have too many in one index, your memory could blow up. So if you have Elastic or Splunk's professional services, or you're hosting it on their cloud, that just makes those types of issues a lot simpler.

Jon McLachlan

Yeah, it makes life easier. It allows us to focus on our business value add, instead of having to worry about a huge tech stack of random services that some open source project just referenced, and now you're deploying. No, that makes perfect sense. I'm right there with you. Almost always I'm for a buy versus build, because I'm the expert of very little in this world and would love to include other people's hard work and time well spent grappling with those important nitty-gritty questions that are all solved underneath the hood of a managed service, so that I can not only take advantage of all of their expertise, but not have to worry about it. It almost seems like a no-brainer. And it's very surprising when I see companies actually invest huge amounts of time and money and engineering into building out things that already exist in this world. It's like, you want to reinvent the, well, okay. I mean, I suppose there's some interesting technical problems underneath the hood that need love, but there's also a lot of open technical problems that haven't been solved yet, and rather work on those.

Wesley Belleman

So I would as well.

Jon McLachlan

It makes perfect sense. So speaking of work, you recently transitioned from mostly your military service to now civilian service, with your role as a system engineer in Palo Alto Networks. Of course, you're still part of the California Air National Guard too, part-time, as a cyber warfare operator, which is very cool. But what would you say is the biggest challenge with the private sector that you've noticed so far in your, you've been there for two months.

Wesley Belleman

You know, honestly, I like Palo Alto. Like, Palo is a really great company, so it's hard to say that there's really any challenges, just because it is such a great company. Like, I've heard people, when they leave the military, they say, like, oh, everything here is, like, cutthroat, and, you know, it's all about personal performance and stuff. And I haven't had that at Palo Alto at all. Like, it's very team focused. It feels, you know, it feels like people care about each other. So I definitely don't have any complaints about being in Palo Alto, and I'm really happy. I think the biggest issue for anyone, especially so for my background, right? Growing up with parents who worked for the government in one form or another, more or less their whole careers, and then me going to work for the government after college, and then now leaving that, I think there was just kind of a fear and, definitely, like, a feeling of instability leaving. And so I think that was a little bit...

Jon McLachlan

You're breaking the tradition.

Wesley Belleman

Right, exactly. So I'm just, you know, I'm leaving, because, I mean, the government can tend to be a little bit more stable as far as salaries, as far as the economy. For example, right, like, we were in 2008, during the financial crash, I was living on a military base and had barely felt the impacts of the financial crash. And I'm sure there's going to be some people listening to this who don't want to hear that. They certainly probably felt a lot of impacts due to that significant financial event in 2008. But almost everybody on base, our little isolated world on this military base, was more or less unaffected, because the government just kept giving out the regular paychecks. There's no layoffs or anything like that. So I think some of those scares were definitely a challenge for me leaving, especially being someone who grew up with that crash in mind. They talk a lot about how millennials a lot of times think about the 2008, and that affects their lives, and that's one of the reasons why millennials are making certain financial decisions that people have probably seen in the news. And so that was the same thing for me. Like, I kind of fear instability due to that major event in my life. And even though my interaction with it was basically, it's not affecting my life, but I know people at home who it is affecting. And therefore I felt very safe and secure. And when I was leaving the military, I felt like I'm leaving that safety and security.

Jon McLachlan

Right. Well, good for you for doing what's right for you and breaking the tradition and taking a risk, really. You know, I'm always a big fan of people that are really willing to try something new to see how it fits. You can always go back if it doesn't work out, but just the experience and that diversity of trying something new brings with it a new perspective oftentimes that can be really rewarding. How about an advantage that the private sector has over military?

Wesley Belleman

Right. No, absolutely. So with kind of that lack of security comes a lot of freedom, right? So I'm able to live where I want, as long as I can get a job in that location. I'm able to move laterally. So right now, you know, I'm a system engineer. If I wanted to go be a software developer, as long as I can, you know, build that tool set, and someone is willing to hire me as a software developer, and they think that I either have the potential to do it or I've been able to learn that in my spare time, that's something I can do. And that kind of internal mobility is very difficult within the military. So you can kind of guide your own career a lot more in the private sector than you can in the military. Versus in the military, you kind of get on the train track, and, you know, that's your track, and you have to kind of just go forward on that track. And then basically your control is the pedal, right? Like, how fast do you go along that track? Not can you turn left or right. Versus in the private sector, you can turn left or right if you want to. And a lot of people, you know, like me, who are a little bit nerdy, we like to get hands-on and build stuff. A lot of times we don't want to be on this kind of up or out track. I want to go continue to be engineers into the future, and sometimes the military doesn't let people do that.

Jon McLachlan

Yeah, for sure. The career paths in large companies, they definitely have two tracks associated with them: the IC role, the individual contributors, you know, the engineers, the operations folks. And you can climb the ladder just as well as an IC sticking to engineering, and you become an architect. But yeah, they facilitate that type of growth and that career path really well, because at the end of the day, they want to make sure everyone is challenged just enough to still be learning and growing in the private sector. And that's really that sweet intersection between, you know, you're doing good work, but you're also learning and growing, and so you're trying out new things. You're engaged. And then when you do have your successes, they're real successes. You know, it's not just, like, mundane, same thing over and over and over again. Well, at least it shouldn't. If you find a good company, if you find a good team, that cycle will present itself over and over again within the same context. So I'm happy for you. It sounds like you've got a little bit of that entrepreneurial spirit, where you like to be challenged and engaged. So it'll be exciting for you there, Palo Alto Networks. And, okay, so what about [piece] of the military, culturally speaking, that you wish you saw a little bit more of in the private sector? Is there anything in that space for you that you thought really worked well in the military, and you think it might also work well in the private sector? The private sector is just missing just a pinch?

Wesley Belleman

No, I wouldn't say anything particular. I mean, I'll just say the military is very, just like the private sector, right? It's very diverse. It can depend on where you go. So for instance, when I was at United States Southern Command, and this is something I've talked about before, when I was at the United States Southern Command, that's more of a, so if people aren't familiar, the military is divided up into service branches, which organize, train and equip. So they organize troops, they train them, and then they equip them with different forms of technology. And then there's combatant commands, which actually do, for lack of a better term, war fighting. And United States Southern Command is one of these combatant commands. And so in any kind of combatant command environment, it's more of a deployed environment. It's more operational and active, versus in a service branch environment, it's more of acquisitions or training, et cetera. So when I was at United States Southern Command, it was a lot more operational, but it did feel, everybody felt like a person, like everybody treated me like a person. You know, everyone was very friendly with me. It wasn't very robotic. People weren't just coming in to just, you know, do their work and get home. It was a lot more, you know, we're all friends, and we're here to be friends. We're here to grow relationships, and we're here to help each other out. And that's probably something that some companies need, that kind of attitude. I'm sure there's a lot of, from what I've heard, I mean, I have a lot of friends who work at those types of companies where it's, you know, punch in, do as much work as possible, and then punch out. But there's certain parts of the military that are like that too. So I think, whether it's in the government or in the industry space, every company should just realize that everyone you work with is a person. Treat them like a person, be friendly with them. Don't just treat them like an object, like an input-output kind of function or something like that. Like, just, like, treat them like a person, be kind to them. And I definitely, you know, I saw that more in one job I had in the military than in other jobs. And I see that at Palo Alto Networks, and I know in other companies, they don't have that mentality. So it doesn't really answer your question exactly how you asked it, but I do think that that's something that I'd like to see just across the board.

Jon McLachlan

Yeah, for sure. I think it's super important to see people first as people. And, you know, when I interact with people, I like to assume two things. One, that they bring something to the table, and it's my job to see what that is. And it's going to be something different from what I bring to the table or what anyone else brings to the table. But it's something. You know, they have some set of experiences and values and knowledge that's valuable. And then the other thing that I like to assume, you know, when I interact with people, is that they want to do the right thing. Right? That the context around whatever's going on is driving them, you know, with their set of values and their experiences, that X is the right thing to do, for whatever X happens to be. And that means that when there's a disagreement between people, it's really one of two things that are about ready to happen. A, someone is going to learn something new about how the world works, right? They're going to gain another piece of knowledge or clarify their view of how the world is working. Or, on the other side of that equation, they're going to have an opportunity to see the world from a new perspective, where it's not even a problem anymore. And that's the answer to the question, you know, what's the right thing to do? And then when you adopt a mentality like that, then disagreements become this really interesting thing, because you know something good is going to come out of it at the end. And you don't know if you're going to be on the receiving side or the giving side of the knowledge piece or the perspective piece. But in all four of those cases, I've been on both giving and receiving sides all of the time, and it just makes everything really enjoyable for everyone. And then it's not like, you know, as long as that core respect is there and you can acknowledge that with each other, because we all have super different life experiences. But it's those experiences that shaped us into the person that we are today. We've got to see that in each other and acknowledge, like, our differences, not as baggage, but as strengths. So I've always been fortunate enough to work for companies in the private sector that really embody that. Maybe not perfectly, but pretty much most of the time, that's what everyone is shooting for. So I feel very fortunate to have had those experiences and to have had the type of career that led to valuing our differences. I think that's the core value of diversity too. You just have very, very different experiences across the board, across different cultures, even within our own nation, our own society. Okay, so let's go back to the security stuff for a second here.

Wesley Belleman

Okay. Oh, I was going to say, well, before, actually, I just want to say I really like, in your article, you talk about celebrate disagreement. I just think that's a really good mantra for people. Maybe mantra is not the right word, but just something for people to remember. Because I think sometimes disagreements looked badly, like, looked as bad, and some leaders might say, well, there's too much disagreement here, I can't make a decision. And I don't think that's always the case. Sometimes disagreement is just part of the process. And if we just accept it and celebrate it, then we won't let it hold us back. One thing I like about Palo Alto is one of our core values is disruption. So in the Air Force, the three core values are integrity first, service before self, and excellence in all we do. And those are great core values for sure. But sometimes I felt like those were all very me-centric, right? Like, my integrity needs to be first. I need to give service before myself, which I don't think those things necessarily have to be competing. In excellence in all we do, like, I have to be excellent in everything I do. In Palo Alto, we have some of these things, like disruption, where, you know, it's about sort of the whole system as a whole, like changing the whole system to be innovative, and to use a buzzword, right? And then also collaboration. We have, specifically, collaboration is also one of our core values. So we're encouraged to collaborate and work together across, you know, across the entire company. And so the Space Force is actually setting up its own core values. So right now it started with the three that the Air Force has, temporarily, but we'll see. Maybe it'll have some more modern types of core values, such as collaboration, inclusion and disruption, which are three of the core values of Palo Alto.

Jon McLachlan

No, that would be awesome. You know, but, you know, larger organizations, the larger the organization, the slower it will change. Maybe that's actually a benefit for the Space Force, because it's new, maybe a little bit smaller. I bet they could be more innovative more quickly than some of the other branches of the military. Super interesting. All right. Well, how about I'll put you on the spot a little bit. Yeah, I know we already talked about some predictions that you had, you know, in terms of, is a SOC going to be entirely automated, or could it be entirely automated? And that's a hot topic, and maybe 10 years we'll have a different conversation around that. But do you have any predictions about what might happen with the SOC space, with sort of attack detection? What do you think? What you see as the future? What are you optimistic about that we're going to see within our lifetimes?

Wesley Belleman

No, that's great. And I definitely wanted to reference someone that I followed a lot on this, Anton Chuvakin. I don't know if you've heard that name, but he works at Google, and he was at Gartner for many years. And he's definitely someone, you know, if you're listening to this and you're interested in these topics, he's definitely someone I've learned a lot from. I've read several of his articles and research. But he was the one who talked about the SOC being automated, and how he's actually, his prediction is that, it seems, at least it seems to indicate that he predicts that the SOC could be fully automated at some point in the future. He said it might be 2025, it might be 2045. He didn't seem to give a specific [date]. But he does predict that, and he makes some good arguments for that. So I would say he's definitely swayed me a little bit in that direction. I'm definitely...

Jon McLachlan

Oh, go ahead.

Wesley Belleman

Go ahead.

Jon McLachlan

Oh, no.

Wesley Belleman

Yes. I'm definitely, I mean, I've got, you know, I've got a little bit of, like, optimism for sure. But I think there's also, we're going to see a lot of change, right? So as a lot of companies start collecting more data, they're going to start to see which data sources are more useful and which are less useful. So I do see the SOC, I could see the size of SOCs kind of ballooning and then shrinking back and then growing, kind of up and down, like a little accordion. And so, you know, what do I mean by that? So, I mean, you know, a lot of SOCs might collect, you know, every single endpoint log that they can think of, and they might layer on an endpoint device on there, an endpoint protection tool, such as from Symantec or another company. Palo Alto has one as well. Lots of companies have these endpoint tools. And you could layer all that on and then collect all this data, and then find that a lot of the data you're not really using, or it doesn't have a lot of predictive power as far as actual attack detection. And so I could see that kind of changing, maybe not moving in a linear direction, but going up and down. One market trend I've seen recently that I was pushing a lot in the Space Force. So I have a presentation I gave in the Space Force several times called Nowhere to Hide. And in that, I covered security automation, which is my primary...

Jon McLachlan

Was it K-N-O-W or N-O?

Wesley Belleman

No, it's just N-O, like nowhere. Nowhere to hide.

Jon McLachlan

So [there's nowhere to hide]. You can't run, nowhere to hide.

Wesley Belleman

Right. So the idea is that, you know, we do all these things, right? So, like, you run an encryption company, right? And encryption is very important, right? And if you can do encryption correctly, then an adversary, even if they can see your data, it's encrypted and they're not able to actually read it, for example, right? But the idea of nowhere to hide is, if the adversary is doing something that they're not supposed to be doing, they can't do that undetected, right? They're going to be basically caught and stopped very quickly. And so a huge part of that was security automation. So you see the SOAR platforms taking off, such as Phantom and Palo Alto's Cortex XSOAR. Those two platforms are taking off in the space to increase security automation and to bring all the different data sources together. So that's a huge part of SOAR. These different SOAR platforms have hundreds of integrations with lots of different tools. And for security operations centers that have dozens of tools, they have tool sprawl, and they don't want to log in each individual one. They can bring it all together in a SOAR platform, and then they can start to add playbooks, which are just basically scripts. Some people don't like scripting, they want, you know, visual programming. So you can do that with a playbook, and then you can automate certain actions within a SOC. So that's definitely, we're going to start to see that take off. And as that starts to grow, we're going to see SOC operations just happen faster and faster. And then SOC operators are going to start moving out of what we call tier one activities, which are just very, very basic SOC activities of just opening tickets, closing out, like, really obvious false positives. And they're going to start to focus on a lot more complicated threat detection. So that's one market trend I see, and definitely something I predict in the near future is going to happen. And then the other thing that we're probably going to see happen in the near future is also in the XDR space. So XDR is an interesting concept. Like, some people could probably argue that XDR has been going on for as long as security operations exist, because XDR just means extended detection response. So rather than doing endpoint detection response or network detection response, you're doing it across the endpoints and the network and other devices. So I know you've talked on your podcast about DLP. DLP is also part of the extended detection and response architecture, right? And so bringing all those data sources together, and then doing the detections across the several data sources in a single data object, rather than just trying to do it within each stream. And what's interesting about XDR is a lot of companies are using similar, like, a lot of companies are using Red Hat, right? A lot of companies are using Windows. And so the log sources from those are going to be very similar. And then network traffic, you know, most network traffic is probably TCP/IP, right? So if you can correlate Windows logs with TCP/IP, then those analytics that correlate those two different data sources can be shared across different organizations. So we're seeing in the XDR space is that companies are able to actually have those analytics on their cloud, and then customers are able to send their data to that cloud to be analyzed with proprietary analytics. And then those analytics can send a report back to the company's security operations center to say, okay, we've seen this traffic look suspicious, so go look into it more. And so that actual commoditization of the analytics and selling of the analytics is one of the major new trends we're seeing in the market, and I think that's definitely going to take off. And also that coming with threat intelligence, which we can talk about if you want, threat intelligence becoming more commoditized and being sold more as well.

Jon McLachlan

So what is threat intelligence? It's just the fingerprints that identify specific types of attacks.

Wesley Belleman

Right, yes, exactly. So the idea of threat intelligence, there's sort of two ideas of threat intelligence. So if an attack happens on Palo Alto's network, that adversary is going to want to use that attack on several different networks to see how many they can compromise. So, you know, especially something like ransomware, right? Like, ransomware works best if you can hit as many targets as possible and get them all to pay. So if there's a ransomware attack that Palo Alto Networks is able to detect, and hopefully stop right before anything happens, then we can share what that attack looks like with [Peacemakr], and then you're able to make sure that that doesn't hit your environment.

Jon McLachlan

Yeah, exactly. I would imagine the ransomware folks, you have to break into these networks. And so there's a period where certain networks have been compromised, and they have access to the storage device, they identify all of the backups. It's quite the attack. You know, I would imagine it's very targeted too. It's hard to automate something like that.

Wesley Belleman

Yes, I mean, it can depend. Certain ransomware attacks have been automated. But that's a good point. That's called the dwell time. So that time from which the adversary is in your network to the time that they're detected is known as the dwell time. Maybe you've heard that term before. But part of the point of security automation and just adding additional sensors is to limit the dwell time. So that's really our goal, you know, within the community, security operations community, is to shrink that dwell time as much as possible. Well, I can give you a different example. So the SolarWinds attack, right? So they said that that started in the summer, so around June or July of 2020. And then it was released, the information was released about it, as detected by CrowdStrike, in December of 2020. So it was about five or six months of dwell time. Right. So the idea is to, you know, we're trying to shrink that down to much faster than that. If you look at, so CrowdStrike actually has a report on different adversaries, and a lot of the adversaries have breakout times, which is kind of a niche term, but it's basically the time in which you need to detect an adversary in order to actually stop them. And, like, the fastest adversary that CrowdStrike had on there, the fastest APT, was about two hours. So, you know, so we're trying to get that dwell time from six months in some cases right down to something that's measured in hours or minutes. So there's a lot of work being done in the analytics, security automation, to just really speed that up.

Jon McLachlan

Excellent. Well, I'm excited for the future. It will be more automated. It'll be faster. We'll be more secure. And no small part to folks like you on the front lines, helping think through these hard problems. Wes, it's been an absolute pleasure. Thanks for sharing your wisdom, your experiences.

Wesley Belleman

No, absolutely. Thank you for having me. This was a great experience. Thank you.

Jon McLachlan

Awesome. Well, we'll have to have you back on at some point, and we'll see how those predictions work out.

Wesley Belleman

Yes, absolutely.

FAQ

Questions this episode answers

Who is Wesley Belleman?
Wesley Belleman is a cyber security consultant at Palo Alto Networks, which he joined in May 2021. In the California Air National Guard he has been a lead cyber planner since March 2023 and became commander of a 44-person Cyber Protection Team in April 2026. From 2017 to 2021 he was a project manager in the Air Force, where he owned the requirements for the Space Force's Kraken and led a team at United States Southern Command.
What is a security operations center?
A SOC is where security operators watch an organization's logs and network traffic, compare them against signatures of known attacks, follow up on alerts, and do incident response when an attack worked. Wikipedia describes a SOC as responsible for protecting an organization against threats.
Does a small company need a SOC?
Wes says every organization should have security operations. A small organization can outsource it to a managed security service provider, or MSSP, which centralizes experts who serve several tenants and so cuts costs.
Can a SOC be fully automated?
Not soon, Wes says. Many SOCs had yet to implement automation and were still getting their data in one place. He expected a different conversation in five to ten years, and recalled Anton Chuvakin predicting a fully automated SOC at some point.
Why is security operations a data problem?
Wes says newer books on security operations start with data engineering and data science. Network, host and service data go into one repository, and telling malicious traffic from benign traffic is a binary classification problem.
What are the five decisions in Wesley Belleman's article?
His February 2021 article in Towards Data Science, Classification in Security Operations, lists five classification decisions: whether data is useful for detection, whether an event is an incident, whether the context data is sufficient, whether it's a true incident, and how effective the mitigation was.
What is Security Onion?
Security Onion is a free Linux distribution for threat hunting, security monitoring and log management, developed by Doug Burks in 2008. Wes calls it a starting point for hobbyists: the Suricata and Zeek network sensors feed the Elastic stack, which you view with Kibana.
What is MITRE Caldera used for?
Wes uses it to validate security tools. Caldera is an open source project from MITRE based on ATT&CK. You install its agent on an endpoint, and it runs techniques such as pivoting, command and control and exfiltration, so you can see them in Security Onion and build detections.
Should you build or buy security monitoring?
Wes is a proponent of buy when you can. Running your own Elastic cluster means configuring log rotation and watching storage and memory. Hosted options such as Elastic Cloud, or Splunk's offerings, take that on.
What is SOAR?
Security orchestration, automation and response platforms connect a SOC's tools and run playbooks, which Wes calls basically scripts. His examples are Phantom and Palo Alto's Cortex XSOAR. They move analysts out of tier one work such as closing obvious false positives.
What is XDR?
Extended detection and response runs detection and response across endpoints, the network and other devices. Wes says vendors can share analytics across organizations because many run the same systems. Wikipedia says Nir Zuk of Palo Alto Networks coined the term in 2018.
What is dwell time?
It's the time from when an adversary is in your network to when they're detected. Wes puts SolarWinds at about five or six months, and wants dwell time measured in hours or minutes.
Why did Wesley Belleman leave the military?
He describes archaic HR practices. You get pigeonholed into one career, have little choice in where you live or in moving laterally, and face up or out, which pushes people who want to stay technical into management.
What happened to Wesley Belleman after the episode?
He stayed at Palo Alto Networks, now as a cyber security consultant. In the California Air National Guard he became a lead cyber planner in March 2023, working on Cyber Dawn, Amber Mist and a NATO exercise, and commander of a 44-person Cyber Protection Team in April 2026.
Keep reading on the YSecurity blog
New episodes every other Tuesday

Got a story like this one?

We're always looking for practitioners who've been in the room when it mattered.